Electric machine monitoring method and system

• US 6,496,782 B1
• Filed: 10/30/2000
• Issued: 12/17/2002
• Est. Priority Date: 10/30/2000
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Claus, Bernhard Erich Hermann, Kliman, Gerald Burt, Chbat, Nicolas Wadih
• Primary Examiner(s): Hilten, John S.
• Assistant Examiner(s): SUN, XIUQIN
• Application Number:
  US09/699,089
• Time in Patent Office: 778 Days
• Field of Search: 702/56, 702/34, 618/490, 618/806, 187/29.R, 731/46, 359/191, 364/551.02
• US Class Current: 702/40
• CPC Class Codes: G05B 23/0221 Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods

Abstract

A machine monitoring method includes obtaining signals indicative of machine conditions, machine rotational speed, direction, and load conditions over a segment of time, transforming the obtained signal indicative of machine conditions into a frequency spectrum, identifying low level features of the frequency spectrum, and processing the low level features to obtain an indicator value representative of the machine conditions.

Claims

1. A machine monitoring method comprising:

(a) obtaining signals indicative of machine conditions, machine rotational speed, direction, and load conditions over a segment of time under substantially constant operating conditions;

(b) transforming the obtained signal indicative of machine conditions into a frequency spectrum;

(c) identifying low level features of the frequency spectrum; and

(d) using the signals indicative of machine rotational speed, direction, and load conditions and processing the low level features to obtain an indicator value representative of the machine conditions.

2. The method of claim 1 further comprising

(e) repeating (a) through (d) and constructing a representation of machine condition degradation as a function of time.

3. The method of claim 2 further including using the representation to monitor the rate of change of degradation.

4. The method of claim 2 further including using the representation to estimate time remaining prior to reaching a critical threshold.

5. The method of claim 1 wherein the low level features include overall energy of the frequency spectrum, relative energy of relevant frequency bands within the frequency spectrum, and frequency spectrum peak shape-characterizing features within the relevant frequency bands.

6. The method of claim 5 wherein (c) includes scaling the relevant frequency bands according to the machine rotational speed.

7. The method of claim 1 further including,

prior to (a), learning for use in (d) by

(i) for each of at least two known machine conditions, for each of a number of rotational speed, direction and load combinations, obtaining a signal indicative of machine conditions over a segment of time, transforming the obtained signal indicative of machine conditions into a frequency spectrum, and identifying low level features of the frequency spectrum,

(ii) for each combination of rotational speed, direction and load, determining a low level discriminator function for at least a number of the low level features,

wherein (d) includes using the low level discriminator function to obtain the indicator value.

8. The method of claim 7 wherein, in the event that the signals indicative of rotational speed, direction, and load conditions do not match a respective one of the number of rotational speed, direction, and load combinations, (d) further includes interpolating predetermined low level discriminator functions to identify an appropriate discriminator function.

9. The method of claim 1 further including,

prior to (a), learning for use in (d) by

(i) for each of at least two known machine conditions, for each of a number of rotational speed, direction and load combinations, obtaining signals indicative of machine conditions, rotational speed, direction, and load conditions over a segment of time, transforming the obtained signal indicative of machine conditions into a frequency spectrum, and identifying low level features of the frequency spectrum and low level features comprising the rotational speeds, directions, and loads, and

(ii) determining a low level discriminator function for at least a number of the low level features,

wherein (d) includes using the low level discriminator function to obtain the indicator value.

10. The method of claim 7 wherein the low level discriminator function comprises a non-linear function.

11. The method of claim 7 wherein the low level discriminator function comprises a linear function including at least one low level weighting factor.

12. The method of claim 11 wherein, for each combination of rotational speed, direction and load, determining the low level weighting factors comprises using two sets of signals indicative of two different machine conditions to determine a linear combination of the low level features of the two sets of signals which is an optimal discriminator between the two sets.

13. The method of claim 11 wherein determining the low level discriminator functions comprises using at least three sets of features indicative of at least three different machine conditions to determine a combination of the low level features of the at least three sets of features which is an optimal discriminator.

14. The method of claim 9 wherein the combination is a linear combination or a non-linear combination.

15. The method of claim 1 wherein processing the low level features to obtain an indicator value comprises using a first set of discriminator functions to combine the low level features into a set of high level features, and then using the high level features, the signals indicative of rotational speed, direction, and load conditions, and a second set of discriminator functions to obtain the indicator value.

16. The method of claim 15 wherein each discriminator function of the first set of discriminator functions corresponds to a respective rotational speed, direction, and load condition.

17. The method of claim 15 wherein multiple discriminator functions of the first set of discriminator functions correspond to a respective rotational speed, direction, and load condition.

18. The method of claim 15 wherein the first set of discriminator functions and the second set of discriminator functions comprise sets of linear or non-linear discriminator functions.

19. The method of claim 15 further including modifying the first set of discriminator functions by clustering the discriminator functions of the first set of discriminator functions into a smaller first set of discriminator functions with at least one good discriminator function remaining for each respective rotational speed, direction, and load condition.

20. The method of claim 15, further including, in the event that the signals indicative of rotational speed, direction, and load conditions do not match rotational speed, direction, and load conditions of a predetermined discriminator function, interpolating predetermined discriminator functions to identify an appropriate discriminator function.

21. A machine monitoring method comprising:

(a) obtaining signals indicative of machine conditions, machine rotational speed, direction, and load conditions over a segment of time under substantially constant operating conditions;

(b) transforming the obtained signal indicative of machine conditions into a frequency spectrum;

(c) identifying low level features of the frequency spectrum, the low level features including frequency spectrum peak shape-characterizing features within relevant frequency bands of the frequency spectrum; and

(d) using the signals indicative of machine rotational speed, direction, and load conditions and processing the low level features to obtain an indicator value representative of the machine bearing wear and gear wear conditions.

22. The method of claim 21 wherein (d) includes using the signals indicative of machine rotational speed, direction, and load conditions to select a discriminator function, and applying the low level features to the selected discriminator function to obtain the indicator value.

23. The method of claim 21 wherein (d) includes using applying the low level features and the signals indicative of machine rotational speed, direction, and load conditions to a discriminator function to obtain the indicator value.

24. The method of claim 21 wherein (d) comprises using a first set of discriminator functions to combine the low level features into a set of high level features, and then using the high level features, the signals indicative of rotational speed, direction, and load conditions, and a second set of discriminator functions to obtain the indicator value.

25. A machine monitoring system comprising:

(a) sensors for obtaining signals indicative of machine conditions, machine rotational speed, direction, and load conditions over a segment of time under substantially constant operating conditions; and

(b) a computer for, transforming the obtained signal indicative of machine conditions into a frequency spectrum, identifying low level features of the frequency spectrum, using the signals indicative of machine rotational speed, direction, and load conditions and processing the low level features to obtain an indicator value representative of the machine conditions.

26. The system of claim 25 wherein the low level features include overall energy of the frequency spectrum, relative energy of relevant frequency bands within the frequency spectrum, and frequency spectrum peak shape-characterizing features within the relevant frequency bands.

27. The system of claim 25 wherein the computer is adapted to process the low level features by applying the signals indicative of machine rotational speed, direction, and load conditions, and the low level features to a discriminator function to obtain the indicator value.

28. The system of claim 25 wherein the computer is adapted to process the low level features by using the signals indicative of machine rotational speed, direction, and load conditions to select a discriminator function, and applying the low level features to the selected discriminator function to obtain the indicator value.

29. The system of claim 28 wherein the discriminator function comprises a non-linear function.

30. The system of claim 28 wherein the low level discriminator function comprises a linear function including at least one low level weighting factor.

31. The system of claim 25 wherein the computer is adapted to process the low level features by using a first set of discriminator functions to combine the low level features into a set of high level features, and then using the high level features, the signals indicative of rotational speed, direction, and load conditions, and a second set of discriminator functions to obtain the indicator value.

32. The system of claim 31 wherein each discriminator function of the first set of discriminator functions corresponds to a respective rotational speed, direction, and load condition.

33. The system of claim 31 wherein multiple discriminator functions of the first set of discriminator functions correspond to a respective rotational speed, direction, and load condition.

34. The system of claim 31 wherein the computer is further adapted to modify the first set of discriminator functions by clustering the discriminator functions of the first set of discriminator functions into a smaller first set of discriminator functions.

35. A machine monitoring system comprising:

(a) means for obtaining signals indicative of machine conditions, machine rotational speed, direction, and load conditions over a segment of time under substantially constant operating conditions;

(b) means for transforming the obtained signal indicative of machine conditions into a frequency spectrum;

(c) means for identifying low level features of the frequency spectrum; and

(d) means for using the signals indicative of machine rotational speed, direction, and load conditions and means for processing the low level features to obtain an indicator value representative of the machine conditions.

36. The system of claim 35 further including,

means for learning for use in (d) by

(i) for each of at least two known machine conditions, for each of a number of rotational speed, direction and load combinations, obtaining at a signal indicative of machine conditions over a segment of time, transforming the obtained signal indicative of machine conditions into a frequency spectrum, and identifying low level features of the frequency spectrum

(ii) for each combination of rotational speed, direction and load, determining a low level discriminator function for at least a number of the low level features,

wherein (d) includes means for using the low level discriminator function to obtain the indicator value.

37. The system of claim 35 further including,

means for learning for use in (d) by

(i) for each of at least two known machine conditions, for each of a number of rotational speed, direction and load combinations, obtaining signals indicative of machine conditions, rotational speed, direction, and load conditions over a segment of time, transforming the obtained signal indicative of machine conditions into a frequency spectrum, and identifying low level features of the frequency spectrum and low level features comprising the rotational speeds, directions, and loads, and

(ii) determining a low level discriminator function for at least a number of the low level features,

wherein (d) includes means for using the low level discriminator function to obtain the indicator value.

38. The system of claim 35 wherein (d) includes means for using a first set of discriminator functions to combine the low level features into a set of high level features, and then using the high level features, the signals indicative of rotational speed, direction, and load conditions, and a second set of discriminator functions to obtain the indicator value.